GA4 puts landing pages in an other row: check cardinality
Diagnose GA4 landing-page reports with an other row, identify unnecessary URL variation, and compare useful content groups without inventing hidden page results.
By Crawlito · 5 min read
In short
- High-cardinality dimensions can cause less common values to be grouped into an other row.
- Random URL values can split one content page into many reporting values.
- Fix unnecessary collection variation while preserving real content distinctions.
On this page
If GA4 groups landing pages into an other row, inspect the number of distinct dimension values and the report's data-quality information. Random query values, identifiers, or unnecessary combinations can make one content page appear as many values. Reduce unintended variation at its source and compare appropriate content groups; do not assign the grouped outcomes to guessed pages.
This is different from a page failing to load. The visitors may have been measured, while the report does not show every less-common dimension value separately.
Record the report definition
Save the property, date range, dimensions, metrics, filters, and data-quality message. Confirm whether the report uses a page path, full location, or landing page with query string.
Google's cardinality guide defines cardinality as the number of unique values for a dimension and explains how high-cardinality dimensions can increase the chance of grouped rows.
Its other-row guide describes how less-common values can be condensed when the relevant table reaches its limits.
Do not interpret other as a literal URL or a channel name. It represents grouped data in that report context.
Keep the selected dimensions visible in the report note. Adding another breakdown can change the number of combinations even when the underlying pages are unchanged.
Inspect the URL variation you collect
Export a representative set of visible page values and inspect the differences. Look for session-like tokens, timestamps, search text, tracking values, or inconsistent host and path formatting.
Do not remove all query parameters by default. Some select genuinely different public content, such as a language or product variant.
Create a classification table:
| Variation | Meaning to review |
|---|---|
| Tracking-only value | Same public content with acquisition context |
| Content selector | Different intended page state |
| Random request token | Potential unnecessary reporting variation |
| Private or personal value | Data collection issue to address |
| Host or slash inconsistency | URL normalization question |
Keep sensitive values out of shared exports. The investigation needs the shape and source of variation, not private visitor data.
The URL parameter guide helps distinguish content identity from tracking before changing collection.
Trace unnecessary values to their source
Find the link builder, router, form, or tag configuration that adds the varying value. A public article URL may be stable while a measurement override appends a new token on every event.
Compare the browser address with the collected page-location field. If they differ, inspect the code that overrides the measurement value.
Do not repair the report by deleting meaningful content distinctions that the product actually uses. A language selector and a random cache-busting token should not be treated the same way.
For generated links, check whether the parameter is needed for the reader's task. If it exists only because a temporary debugging mechanism escaped into production, remove it through the owning workflow.
Keep acquisition labels according to the measurement plan. The goal is accurate collection, not making all visits look identical.
Record one example from the source code or configuration and its corresponding collected value to prove where the unnecessary variation enters.
Choose a useful reporting level
If the question is which guide leads to completed checks, a content-level grouping may be more useful than every full URL variation. Define the grouping explicitly.
Use a stable content identifier or a deliberate normalized page dimension where supported by the reporting design. Do not silently rewrite values in a way that combines unrelated resources.
Keep separate views for questions that need the variant detail. A product team investigating language-specific outcomes may need a finer breakdown than a content editor ranking article topics.
The content inventory guide helps map public pages to distinct questions. That inventory can inform a useful reporting grouping without inventing the data hidden inside other.
Label the denominator and grouping in the report. A content-level conversion rate and a full-URL row count are not interchangeable metrics.
Verify the collection repair prospectively
After changing unnecessary URL generation or measurement overrides, run a controlled visit and inspect the collected fields. Confirm the intended stable value is used.
Test a genuine content variant as a control. The repair should preserve its meaningful distinction rather than collapsing everything to the homepage or one path.
Also test the completed outcome event. A clean page dimension is not enough if the event represents a button click instead of a successful action.
The conversion measurement guide explains that event-definition boundary.
Record the change date. Correcting future collection does not automatically rewrite historical dimension values or expand every previously grouped report row.
Use the platform's supported reporting tools to assess the new period, and keep any remaining data-quality indicators visible.
Avoid inventing the hidden distribution
An other row's total does not tell you which omitted page produced each visit or outcome. Do not distribute it across visible pages in proportion to their traffic unless you explicitly present that as an unsupported assumption, which is usually unhelpful for content decisions.
If a different supported report or export provides more suitable detail, use it with its own limits and scope. Do not imply that every reporting surface has identical aggregation behavior.
A useful conclusion might be that a particular parameter creates unnecessary unique values and the collection has been corrected. That is actionable even before you can compare a complete healthy period.
Do not call a large other row evidence that SEO traffic disappeared. Separate collection, aggregation, and actual visitor behavior.
For agencies, show the report's warning and the chosen alternative level of analysis so the client can assess the practical limits.
Keep the measurement vocabulary small and meaningful
Review new custom dimensions before launch. A value that is unique for every event is usually unsuitable as a routine reporting breakdown unless the business question genuinely needs it.
Keep event-level identifiers in the systems designed for that purpose rather than promoting every identifier into a dashboard dimension.
When a new campaign or app feature changes URLs, check whether it creates many unnecessary public page values. This catches the issue at the source rather than after a report becomes difficult to use.
Document the meaning of each normalized content group and who maintains it. A grouping that drifts as pages move can create its own measurement errors.
For a release sample, include a normal article, a tracking-only variation, and a meaningful content variant. Their expected reporting relationship should be deliberate and testable.
As a diagnostic comparison, remove an optional breakdown from a copy of the report while preserving the original. Record whether the grouped row changes and keep the two report definitions attached to the result. A landing-page report broken down by another highly variable dimension asks for more combinations than the page-only view. This comparison can identify a reporting-design contribution without proving that any underlying visit was lost or recovered.
Frequently asked questions
Does other mean the visitor's landing page was unknown?
Not necessarily. It can reflect report aggregation of less-common dimension values rather than absence of the original collection signal.
Should I remove every query string from analytics?
No. Preserve meaningful distinctions and the approved attribution design while removing unnecessary variation.
Will a collection change repair old reports automatically?
Do not assume so. Mark the change date and distinguish historical reporting from corrected future collection.
Use the free Crawlito check on a priority content URL while keeping its technical state separate from reporting aggregation.
Filed under Strategy and audits. Topics: analytics, url-parameters, audits